labs pacoima calif unclassified s!m/ieeeeeeee ...2. x-ray diffraction (xrd) 12 3. scanning electron...
TRANSCRIPT
AD-AO" M2 SAN FERNANDO LABS PACOIMA CALIF F/6 13/AINVESTIGATION OF CNTD MECHANISM AND ITS EFFECT ON MICROSTRUCTI*-ETC(U)OCT 80 D 6 SAT N000 978-C-0557
UNCLASSIFIEDS!m/IEEEEEEEEEmlhlllhlllllluEli/IEl/EE-ElIII/EEIEBIB///hE
EE--EmE~hIE-ElllllEllEEI
i lll/EUU ,.k T
11136
II.2L5 NIhI~ .
TiwuYTGAZICU OF CUTD NBCONAIM AN1D ITS J
' 6~vr aPUC Mb UCI)CURAL fl.OPIIS
ICOM NTR O* R0019-78-C-0557 L V
C1 Aeport Prepared for:
U.S. Naval Air Systems Cousand
* Wahington, D. C.
wE Report Period - 9/20/78 to 9/20/79 SJ N 811
S.S
* T Prepared By :Deepak G. U1hat, Ph.D.Research Engineer
ISAN FERNANDO LBRME
Pacoim, Calif ornia
C ~AW~W"e ift p*ilc welease; &uituio uiadate
T
J IVESTIGATION OF CNTD MECHANISM AND ITS
EFFECT ON MICROSTRUCTURAL PROPERTIES,
CONTRACT NO.- NOO019-78-C-0557
Report Prepared for:,/j)//
U.S. Naval Air Systems Command
Washington, D.C.
Report Period - 9/20/78 to 9/20/79
,, 1 , K/, ,,/ * t' 2 -"/ '/ - / * !: X /
Prepared By :,0teepak G./Bhati Ph.D.Research--Eng4i r
Accession For
NTIS GRA&IDTIC TAB
SAN FERNANDO LABORATORIES Iu t Un c t iO l
J.,-tification
Pacoima, California
ir~t>,. '
-. 0'.
Approved for public release; distribution unlimited
October, 1980.
Dlst : . .. 4,
SECURITY CLASSIFICATION OF THIS PAGE (Wten Dota ;nterOd)a
REPORT DOCUMENTATION PAGE BEFORE COMPLETING FSU O1. REPORT NUMBER 2. GOVT ACCESSION NO. 3. RECIPIENT'S CATALOG NUMBER
4. TITLE (and Subtitle) 5. TYPE OF REPORT & PERIOD COVERED
INVESTIGATION OF THE CNTD MECHANISM AND ITS
EFFECT ON THE MICROSTRUCTURE AND PROPERTIES 9/20/78 to 9/20/79
OF SILICON NITRIDE. S. PERFORMING OG. REPORT NUMBER
7. AUTHOR(a) 6. CONTRACT OR GRANT NUMBER(O)
D.G. Bhat N00019-78-C-0557Research Engineer
9. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT. PROJECT. TASK
AREA & WORK UNIT NUMBERS
San Fernando Laboratories - "
10258 Norris Ave.Patinma CA Q1 31
I1. CONTROLLING OFFICE NAME AND ADDRESS 12. REPORT DATE
NAVAL AIR SYSTEMS COMMAND October, 1980
Washington, D.C. 20361 13. NUMBER OF PAGES
7414. MONITORING AGENCY NAME & ADDRESS(if different from Controlling Office) 15. SECURITY CLASS. (of this report)
DCASMA Van Nuys UNCLASSIFIED6230 Van Nuys Blvd. I5a. DECLASSIFICATION/ DOWNGRADING
Van Nuys, California 91408 SCHEDULE
IS. DISTRIBUTION STATEMENT (of this Report)
17. DISTRIBUTION STATEMENT (of the abstract entered in Block 20, iI different from Report)
1. SUPPLEMENTARY NOTES
IS. KEY WORDS (Continue on revere side if necessary and identify by block number)
Silicon NitrideChemical Vapor DepositionCVDGrain Refinement
20. ABSTRACT (Continue an reverse aide If necessary and identify by block number)
-- This report presents results of a research program in which we sought todevelop a chemical vapor deposition (CVD) method for the deposition ofextremely fine grained silicon nitride.
The program consisted of three separate technical efforts. The firsteffort, a parametric study of the conventional silicon tetrahalide- moniaCVD chemistry, produced no significant grain refinement in the Si3N4 deposits..
D Or 1473 EDITION OF I NOV 65 IS OBSOLETE
SECURITY CLASSIFICATION OF THIS PAGE (WMen Dese &ntNe)
sacuIiTy CLASSIFICATION OF THIS PAG[(Wlhm Daa fntede)
-The second effort attempted, with no success, to utilize silicon halidedisproportionation chemistr the CVD process. Finally, we observedan apparently successful S13NV 4 grain refinement during the third effortin which we used the competing codeposition of separate phases to interruptgrain growth. During this effort, we tried the codeposition of siliconnitride and silicon carbide with no success. However, we found apparentlygood results when silicon nitride was codeposited with aluminum nitride. /
/
H
SEcuI'v CLASSIICAIOW OPr
* AGIv'en flafa Er,
TABLE OF CONTENTS PAGE
ACKNOWLEDGEMENTS i
LIST OF FIGURES ii
LIST OF TABLES iv
I INTRODUCTION 1
II EXPERIMENTAL PROCEDURE 3
(a) Deposition of Si 3 N4 3
1. Parametric Study 3
2. Silicon Deposition by Disproportionation of a Subhalide 6
3. "Alloying" of Si3N4 for Grain Refinement 8
4. Deposition of Test Bars 11
(b) Evaluation of Si3N4 Deposits 11
1. Transverse Rupture Strength (TRS) Measurements 11
2. X-Ray Diffraction (XRD) 12
3. Scanning Electron Microscopy (SEM) and Energy DispersiveAnalysis of X-rays (EDAX) 12
4. Electron Probe Microanalysis (EPMA) 13
5. Hardness and Fracture Toughness 13
6. Electrical Properties 13
III. RESULTS AND DISCUSSION 15
(a) Parametric Study 15
(b) Silicon Deposition by Disproportionation of a Subhalide 28
(c) "Alloying" of Si3N4 for Grain Refinement 40
(d) Deposition of Si3N4 on Bend Bar Specimens 51
(e) Measurement of Electrical Properties 57
IV. SUMMARY AND CONCLUSIONS 58
References 61
Appendix I: Deposition Conditions for Si3N4 63
Appendix II: Calculation of Transverse Rupture Strength of a 70Coated Beam.
ACKNOWLEDGEMENTS
The work reported here was performed under Contract
#NO0019-78-C-0557 for the Department of the Navy, Naval
Air Systems Command, Washington, D.C.
The author is grateful to Robert A. Holzl, President,
San Fernando Laboratories for the initial guidance of the
research, encouragement and numerous valuable discussions.
Thanks are also due to Drs. Jacob Stiglich and Rodney M. Panos
for valuable discussions and suggestions. The deposition
work was performed by Messrs. Sam Rustomji, Gerald Galarneau,
Clifford Lewis and Philip Kalkowski under the expert super-
vision of Benjamin Tilley. Ms. Colleen Murphy assisted in
the compilation of experimental data.
i
LIST OF FIGURESPAGE
1. Schematic of Deposition Chamber for Silicon Nitride 5Deposition in a Furnace.
2. Schematic of Deposition Chamber for Silicon Halide 7Disproportionation study.
3. Schematic of Deposition Chamber for Aluminum 10"Dopant" Study.
4. Morphology of Si3N4 Made with Nitrogen as the 19
carrier gas.
5. Various crystal morphologies of Si3N4 with Argon 22as the Diluent Gas.
6. Morphology of Si3N4 made with SiF 4 as Silicon Source. 27
7. Free Energy of Formation as a Function of Temperature 30for Various Reactions.
8. Equilibrium Degree of Completion of SiCl4 Reduction 31as a Function of Temperature, Pressure and Degree ofDilution.
9. Equilibrium Degree of Completion of SiHCl Reduction as 32a Function of Temperature, Pressure and DRgree of Dilution.
10. Morphology of Deposit Made in the Silicon Halide 37Disproportionation Study with SiCl4.
A) 500/2500X
B) 600/3000X
11. Morphology of Deposit Made in the Silicon Halide 39Disproportionation Study with SiHCl 3.
A) 70X
B) 200X/1O00X
12. Morphology of Si3N4 Made with Additions of Propane 41to the gas stream.
A) 1000X/5000X
B) 1000X
ii
LIST OF FIGURES (Continued)
PAGE
13. Morphology of Si3N4 made with Methyltrichlorosilane 44and ammonia.
A) AGPP = 0.61 torr 200X/10OOX
B) AGPP = 1.7 torr 200X/O00OX
14. Morphology of Si3N4 made (A) Without and (B) with 49Al addition.
15. X-ray Elemental Density Maps for the Sample in 50Figure 14(b) Showing the Distribution of (A) Aland (B) Si.
16. Morphology of Si3 N4 Deposited in the Aluminum Dopant 52Study. (A) 2000X (B) 50OX (C) lOOX (D) 1000X
17. Morphology of Si N 4 Deposit on Bend Bar Specimens. 56(A) 2000X (B 2000X
ifi
LIST OF TABLES
PAGE
1. Operating conditions for obtaining microprobe data. 14
2. Summary of results on Si3N4 made with nitrogen as the 17
carrier gas.
3. Summary of results on Si3N4 made with argon as the 21
diluent gas.
4. Summary of results on Si3N 4 made with SiF 4 as the silicon 26
source.
5. Summary of results on deposits made in the study of 36SiC12 disproportionation.
6. Summary of results on Si3N4 made with additions of 42propane.
7. Summary of results on Si3N 4 made with CHjSiCl3 as the 46
silicon source.
8. Summary of results on S13N4 made with aluminum addition. 47
9. Summary of results on Si3N4 deposited on bend bar 55specimens.
10. Dielectric Properties of Si3A4 at room temperature. 60
iv
I. INTRODUCTION
The research and development group at San Fernando Laboratories
has been engaged in an in-depth study of the characteristics of
chemically vapor deposited silicon nitride for the past several years.
A major effort in this regard was sponsored by the Department of
the Navy (NAVAIR). A summary report, covering the activities of the
first year which ended in July, 1978 was written under Contract No.
NO0019-77-C-0557. In the present report, we describe the results
of the development work carried out in thL second yuar of the program
under Contract No. N00019-78-C-0557. This work concluded in
September, 1979.
During the first year of effort, we concentrated on the study of
various process parameters in the SiCl,,/NH 3/H2 system. The objective
was to define the deposition parameters that would result in a fine-
grained, dense deposit of a-Si 3N4 on resistively heated tungsten
filaments. We discovered that the substrate temperature and the
total pressure were the major variables which influenced the morphology
of the deposit. We also found that it was possible to influence the
morphology and grain size of the deposits by controlled additions
of hydrocarbons to the gas stream. The measurements of flexure strength,
hardness and fracture toughness (by the indentation technique 2 ) yielded
average values of 550 MPa (80 ksi), 2500-3500 HV5 0 0 and 3-5 MPaVm'
respectively. Isolated values of strength and fracture toughness of
1000 MPa (145 ksi) and 7 MPaAm' suggested the potential of this material
that could be realized by a better understanding and control of the
~-1-
process parameters. We recommended that the efforts be continued
to achieve this understanding and control, and also to attempt the
deposition of translucent or transparent Si3N4 for possible use in
electro-optical applications.
Thus, the objective of the effort during the second year was to carry
out an extended parametric study of the silicon nitride deposition.
Ultimately, we hoped to apply the technique of controlled nucleation
thermochemical deposition (CNTD)* 3-5 to this material. Essentially,
the CNTD process results in a deposit of extremely fine grain size,
of the order of 500-l00OR, and superior mechanical properties. This
process has been successfully applied to the tungsten-carbon system 3
and silicon carbide. 4-6 Other systems in which limited success was
achieved in the application of CNTD include Ti-B and Zr-B . The
significant success with the CNTD process in the SiC system 6
prompted us to examine the possibility of extending the technique to
the other Si based ceramic systems, such as Si3N4. As mentioned earlier,
the efforts during the second year of the NAVAIR program were, therefore,
directed towards this objective. These efforts are described in the
following pages.
We divided the experimental work into a number of categories. In the
first phase, efforts were made to establish process parameters under
the condititions of indirect heating of the substrates in a furnace.
Several variables were selected for study so as to define a set of
conditions for the optimum deposition of Si3N4 with a given gas mixture.
*Process developed and patented by San Fernando Laboratories, a division
of Dart Industries.
-2-
The second phase of the program was onducted concurrently with
a similar program on the development of silicon carbide under the
auspices of Air Force Office of Scientific Research (AFOSR), This work involved
deposition of elemental silicon by the disproportionation of a lower
halide and subsequently, conversion of silicon to SiC or Si3N4 using
appropriate source.
In the third phase of this program, we attempted to co-deposit SiC
to achieve grain refinement. We also studied the effect of "alloying"
of Si3N4 by other compatible cations such as Al. It was expected that
by setting up competitive reactions, it might be possible to prevent
unilateral, columnar growth of any one specie, thereby effecting
grain refinement. In the final stage of the program, several test bars were
coated with Si3N4 for detailed evaluation of structure and properties.I
II. EXPERIMENTAL PROCEDURE:
(a) Deposition of Si3 N4.
The outline of the experimental effort for the second
year was based primarily on the experience gained during the first
i - years' work. Several goals were defined, as described below.
1. Parametric Study:
The first objective was to change the method of heating
the substrate. During the first year, we used tungsten filaments
which were heated by internal resistance in a "cold-wall" reactor. We
decided to use graphite bend-bar type substrates which would be
heated indirectly in a furnace. The advantages of the latter type
(3)
of arrangement are (i) easier scale up (ii) possibility of depositing
on complex shapes, and (iii) no restrictions with regard to electrical
conductivity. Thus, we modified the design of the reactor chamber to
allow for the indirect heating of graphite substrate. Figure 1 shows
the schematic arrangement of the deposition chamber. The graphite
furnace was heated by induction by coils placed around the quartz
envelope surrounding the furnace. A clamshell-type heater was incorpor-
ated on the upstream side to permit preheating of the gas stream.
This arrangement was used to deposit conventional silicon nitride on
the bend bar substrates, during the initial parametric study.
The parametric study was divided into several sets of experiments. These
were (i) use of nitrogen as the carrier gas for SiC1 4 and NH 3, (ii) use
of argon as the diluent gas with no nitrogren in the gas stream, and
(iii) use of SiF 4 as the source of Si.
In the first set, viz. nitrogen as carrier gas for SiCI4 and NH3 , a
total of 32 runs were made ±n which the effect of various parameters was
studied with respect to the rate of deposition and morphology of the
crystallites deposited. The run conditions are given in Table A-1 of
Appendix I. We examined the nature of Si3N4 deposits as a function of
substrate temperature, total pressure, active gas partial pressure (AGPP),
partial pressure of hydrogen and the throughput velocity of the gases
at a constant ratio of SiCl4 to NH 3 of 0.2 (except run #29, see Table A-1
Appendix I). The active gas partial pressure was calculated according to
the stoichiometric proportion of the two species required to make a mole
-4-
NH3 -Ip- 2
SiC14 + N2Chamber Pressure
QuartzChambers
SpecimensIpacetrs
TouVnccue
SpciecienTI
DEPOSITIO INnAuFRNACE
-5-
of Si3N4 -. For example, in the reaction,
3 SiCl4 + 4 NH3 -, Si3N4 4 12HCl (1)
we need 4 moles of NH3 for every 3 moles of SiCIk to make a mole
of Si3N4 . Thus the active gas concentration was obtained by adding
the volumes of SiCl4 and NH3 in the proportion 3:4. Any excess of
SiC1 or NH3 was treated as such, and not included in the calculation.
In the second set of experiments, argon was used as a diluent gas. The
variables were total pressure, substrate temperature, SiCl4/NH 3 ratio,
AGPP, partial pressure of hydrogen and the throughput velocity. A total
of 15 runs were made. The run conditions are given in Table A-2,
Appendix I.
Another useful source of silicon is SiF4. Several runs were made with
this precursor to study the effect of AGPP on the rate of deposition and
properties of the deposit. The run conditions are summarized in
Table A-3, Appendix I.
2. Silicon deposition by disproportionation of a subhalide:
For the second part of the effort, we used various
methods for the deposition of elemental silicon. The reactor chamber
was modified to accommodate a smaller chamber in which silicon bearing
solid materials could be placed. The arrangement is shown in the
schematic of Figure 2. The graphite pot, placed over the furnace
-6-
I
Hi NH3 SiCl4 + N2/Ar
Graphite Pot
TIC
Quartz
Molbydenum Tube
Graphite Potfor Si3N4 chips
J - S3N4 chips
Graphite Furnace I ction Coils
Specimens
To
Vacuum
S,.--Motor Drive for
Specimen T/C fSpecimen Rotation
FIGURE 2 SCHEMATIC OF DEPOSITION CHAMBER FOR SILICON HALIDE
6-v DISPROPORTIONATION STUDY.
was used to hold semiconductor grade silicon chips or silicon nitride
chips and scrap. The silicon bearing precursor gas was passed over
this material to achieve the formation of a lower halide, which
was subsequently disproportionated over the substrate in the furnace.
The initial experiments involving 12 runs were made under the AFOSR
program. SiCI4 was passed over hot silicon nitride scrap (mostly
RBSN from different sources) and then allowed to enter the furnace
to react with ammonia. These runs were then continued under the
present contract and an additional 34 runs were made. In the last three
runs, SiHC13 was used as the precursor gas in place of SiCl4 . The run
conditions are summarized in Table A-4, Appendix I.
3. "Alloying" of SijN4 for grain refinement:
This phase of the program was aimed at grain refinement
by alloying. In this work, we first examined the effect of adding
propane to the gas stream. The purpose was to study the possibility
of carbonitriding of silicon and thereby attempt grain refinement
in the deposit. Most of the nine runs were carried out at a constant
temperature of 1375°C and a total pressure of 40 torr. The variables
were hydrogen pressure, velocity of gases and amount of propane. In
calculating the hydrogen pressure, the contribution of propane
(4 moles of H,, for each mole of propane) was taken into account. The
method used in calculating the active gas partial pressure is discussed
in the next section. The run conditions are given in Table A-5, Appendix I.
I'1
In an alternative approach to the same goal, four runs were made
with methyltrichlorosilane (MTS) as a source of silicon. Again, the
purpose of this brief set of runs was to examine the possibility of
co-depositing SiC and SijN4, since MTS is used for the deposition of
SiC. We attempted a quick survey of the effect of substrate temperature,
hydrogen pressure and total flow on the nature of the deposit. As
discussed in the next section, the results were not encouraging,
therefore, no further work was done. The run conditions are given
in Table A-6, Appendix I.
We were examining the possibility of refining the grain structure of
AIN by the introduction of Si on a concurrent program for Al
development under the auspices of AFOSR. 13 These experiments gave
encouraging results for AIN. Therefore we attempted to carry out
similar experiments for introducing Al into SiN 4 . While attempting
to effect "alloying" of AIN by Si, we had made a brief attempt to
do the same at the other end, i.e. "alloying" of Si3N4 by Al, These
runs, included in this report (see Table A-7, Appendix I), gave
encouraging results. Therefore, we continued this effort under this
program.
The furnace design for these experiments is shown in Figure 3. The
inner quartz chamber was used for aluminum granules. The chamber was
heated by a clamshell heater placed round the outer quartz envelope.
Aluminum was converted to AlCl by reacting with HC1, and then
introduced into the main gas stream near the furnace below. The deposition
O-9-
4 + H2 +-N Chamber pressure
Quartz chambers
Clamshell heater
-Aluminum granules
Graphite Furnace
3bInduction Coils
Specimens
Motor Drive forSpecimen Rotation
Specimen T/C
FIGURE 3 SCHEMATIC OF DEPOSITION CHAMBER FOR ALUMINUM "DOPANT" STUDY
parameters of these experiments are given in Table A-7, Appendix I.
4. Deposition of test bars:
Several test specimens were made by depositing Si3Nqon
reaction bonded silicon nitride substrates. These samples were submitted
to NAVAIR for evaluation and testing. The deposition conditions for
these bars are shown in Table A-8, Appendix I.
(b) Evaluation of Si3 N 4 deposits.
The deposits of Si3N4 made in the various runs were
characterized by different techniques. We used mechanical testing,
microscopy (optical and SEM), X-ray diffraction and electron microprobe
analysis. A brief description of the apparatus and procedure follows.
1. Transverse Rupture Strength (TRS) measurement
A table model mechanical testing machine made by Comten
Corporation, St. Petersburg, Florida was used. The platen of the
machine was fixed to a screw driven by a motor through a reduction gear
train. The linear displacement rate of the machine platen was
1.27mm/min. Flexure testing was carried out in a three-point config-
uration. The fixture incorporated sintered tungsten carbide loading
pins. The load at failure was displayed on a mechanical force gauge.
We used mechanical force gauges (capacities 0-100 lbs. and 0-250 lbs.)
instead of the hydraulic load cells since the former were more accurate
and precise in the range of fracture loads encountered with deposits on
graphite substrates. The standard test specimens were prepared by
1" -11-
A
depositing Si3N4 on graphite bend bars with dimensions 0.1" x 0.2"
3.0" (nominal). Some RBSN bend bars, 0.125" x 0.25" x 3.0", coated
with Si3N4 were also tested. The method of computation of the strength
is given in Appendix II.
2. X-ray diffraction (XRD):
The crystallographic identification of the deposits was
carried out on a General Electric XRD-5 unit at the University of
Southern California. A nickel-filtered copper Ka radiation was used
in all experiments, along with a 30 diverging slit and a medium
resolution 0.20 receiving slit (Soller). Most of the work was
done on as-deposited specimens, except in the case of very coarse-
grained, rough deposits. These were diamond-ground to a flat finish
using 70 micron grit. We also used some samples in a crushed powder
form by grinding some deposit layers in an agate mortar. The samples
were scanned through 150 to 900 20 and the patterns compared with
those from ASTM card #9-250 of the Powder Diffraction File and with
a computer-generated pattern developed by Gazzara and Reed. 8
3. Scanning Electron Microscopy (SEM) and Energy Dispersive
Analysis of X-rays (EDAX):
The electron microscopy work was performed on the
deposits using an AMR 1200A SEM in our metallurgical laboratory.
In addition, we used a Cambridge Stereoscan S4-10 SEM at the University
of Southern California. The SEM examination was limited to an
observation of as-deposited and fracture surfaces. Etched cross
sections were not examined since we have not found a suitable room
-12-
temperature etchant for pure silicon nitride. A Tracor Northern
EDAX system attached to the Cambridge SEM was used for the analysis
of silicon content of the deposits. This information was semi-quanti-
tative at best since the EDAX method is not amenable to the quantitative
determination of light elements (Z<ll).
4. Electron Probe Microanalysis (EPMA):
This technique provides a means of obtaining a fully
quantitative chemical analysis. Samples from the "alloying" experiments
were subjected to EPMA. The analysis was periormed on an ETEC Rl
SEM equipped with an Autoscan Crystal Spectrometer, at Scanning Electron
Analysis Laboratories, Los Angeles. The operating conditions used are
shown in Table I.
5. Hardness and Fracture Toughness
These properties were determined with the aid of a Leitz
Miniload Microhardness tester using a Vickers diamond indentor. Hardness
was measured at loads ranging from 100 to 500g. The fracture toughness
was calculated from the measurement of the length of cracks generated
by indentation. This technique is described by Evans and Charles. 2
6. Electrical properties
The material made under this program was not used for the
evaluation of electrical properties. However, these measurements were
performed by an outside agency on silicon nitride deposits made under
another program. Since San . rnando Laboratories did not participate
AU -13-
1-.4
q-4 i- C 4 .44 x x
044 LP~* O -
~ 414.1 U Lfn kn
- 0 00
z4-4 $4
41 %0 '.0 -4 C -
'.. 0 OC
TZ4 0. 04 w
001-4 14
02 0
Q E-4 0-
04. 04L n W.4.
'-4 0 0)
.0o r 0
0b 0 0 0 0 0 w~ 4to.u-40 to -4 . 4 C4 ('4 4 co 4.
0) 41 I0) 0
co 4.
14 -14-41 r4 -
in the evaluation, the details of the techniques were not available.
The results were, however, made available to us through the courtesy of
Mr. Leggett of Hughes Aircraft Company. These are included in the
following section.
III RESULTS AND DISCUSSION
The study of the relationship between deposition parameters and
the characteristics of the deposit was begun on the basis of some
experience gained in the furnace deposition experiments conducted in
a company funded IR&D program. This work was directed towards the
deposition of high strength silicon nitride in thick sections using
graphite substrates. Using the deposition parameters from this study, we
started our effort in the present program. The results of the
various experiments aimed at refining the grain size of the deposits
are described below. The characterization involved examination of
crystal morphology, deposition rate and mechanical properties such
as hardness and fracture toughness. Several specimens were also
tested for transverse rupture strength. The findings of these evalu-
ations are described in the following:
(a) Parametric study:
This study was initiated with a set of parameters in which
nitrogen was used as the carrier gas. Although thermodynamic
calculations suggest that a reaction between SiC14 and N2 in the
presence of hydrogen should yield Si3N4 in the temperature range 1600-
1700K (AGO - -40 to -50 kcal/mol), experience has shown that this does
not happen.
-15-
Kijiima, et al 3 were able to grow 'LSi 3N4 whiskers in the temperature
range 1675-17750K at 1 atm. total pressure when they maintained
a high nitrogen partial pressure (>0.5 atm.) and using very high purity
gases containing less than 10 ppm oxygen and 0.5 ppm H20. At lower
temperatures polycrystalline or amorphous deposits were obtained. In
our experiments, we used nitrogen principally to adjust the throughput
velocity of the gases. Table 2 gives the values of various deposition
parameters in this set of experiments.
The total chamber pressure was varied between 25 torr and about 60 torr.
The low pressure conditions were achieved by connecting the system to
an aspirator. Thus, minor changes in chamber pressure occurred around
a set value depending upon the barometric pressure and ambient temperature
fluctuations. Then, the experiments may be divided into four sets in
which the chamber pressure was maintained in a given range, e.g. 25-30
torr, 39-42 torr, 45-55 torr and 58-62 torr. Within each set, we
examined the effects of other parameters such as the active gas partial
pressure (AGPP), partial pressure of hydrogen and gas velocity on the
deposition rate. The method of calculation of AGPP was described in the
previous section. The velocity of the gases was corrected for chamber
pressure and substrate temperature. Except for one run, the SiCl4/NH 3
ratio was held constant at 0.2.
When the variation of deposition rate was examined as a function of
other parameters within a given set, we could not find any systematic
correlation. An examination of Table 2 shows that the deposit morphology
also appeared to be unrelated to any given parameter.
-16-
FAbLE 2 Summary of results on SIAN, made With nitrogen as carrier gas
GasTotal P* Velocity SlC,, Deposition K TRS
Run Pressure Substrate ACPP H at TP Ni( j rate tlV 200 3 point0 rorr. Temp. OK Torr. Torr M/S Ratio nem/hr _ 0kt MPaW MPa(ksI) Remarks
1-1 25 1b35 0.53 3.0 23.2 0.2 78 Uniform, fine grained deposit.
1-2 25 1575 0.53 3MC 22.3 0.2 78
1-3 25 1500 0.53 3.0 21.3 0.2 - Fine gralned deposit with -omelarge growth spots - clearcrystal lftes.
1-4 25 1500 1.10 6.3 10.6 0.2 - Cuarser grains than prcvlousrun,, some tendency for
wiisker-ilike growth in fewareas.
1-5 26 1650 0.36 2.1 33.8 0.2 48 Clear, fine grained deposit,some spikes.
I-b 39 1650 0.30 1.7 41.8 0.2 48 As above, but no spikes.
1-7 52 1650 0.30 1.7 41.8 0,2 48 Clear, fine grained deposit.
.1 1-8 52 1650 0.31) [.7 41.8 0.2 66 2750 2.95 Clear crystalline deposit-none spikes - Rs substrate.
i-9 48 1620 0.27 1.7 44.5 0.2 48
1-10 64 1650 0.29 1.6 43.8 0.2 36 2800 2.6 Pine crystals and spikcs,cracks on top. RBSN substrate.
1-11 bO 1680 0.27 1.5 47.4 0.2 18 Fine grained clear deposit.
1-12 66 1725 0.30 1.6 44.4 0.2 60 As above, some green colorationIn the 'rystals.
1-13 25 1635 0.53 3.0 23.2 0.2 66 2770 3.35 85.5 Fine crystallites. come
(12.4) spikes. KIISN obist rate.
1-14 40 1650 0.30 1.7 40.8 0.2 102 2725 4.6 152.0 Fine grained atd clear,
(22.0) tranopcrcet de oSI t.eNhS. R -U5 s ,bitrat.
1-15 52 1700 0.30 1.7 43.1 0.2 96
1-16 55 1650 0.32 1.8 39.5 0.2 24 Poor adhesion to sobstroteno spikes, fine depusitL.
1-17 62 1650 0.36 2.0 35.1 0.2 54 Adhesion Is -ltter,d,rk crystals.
1-18 61 1700 0.36 2.0 36.7 0.2 36 l)irk crystalhine deposit ontop, white crystalline ounrest of the hat.
1-19 58 1750 0.33 1.9 39.8 0.2 72 As above.
1-20 58 1800 0.33 1.9 40.9 0.2 - As above, bottom showspoor adhesion.
]-21 59 1650 0.33 1.9 36.8 0.2 - Very thin deposit with a
fine crystallite size.
1-22 45 1650 1.34 2.0 36.2 0.2 66 As above.
12i 47 1650 0.36 2.0 34.7 0.2 - Clear, thin deposit.
1-24 78 1650 0.64 3.7 39.1 0.2 150
1-25 48 1650 0.45 2.6 34.8 0.2 - Thin deposit, partly crystal-line.
1-29 28 1525 1.314 7.3 22.1 2.9 - Poor adhesion of deposit.
1-35 41 1645 0.33 1.9 47.8 0.2 48 Strong (102) orientation, linedark crystalline deposit. [L)AX:('0.5 w/o Si.
1-36 40 1645 0.33 1.8 49.0 0.2 - Coarse(20-2000m)crystallites
on fine grained (mla) matris.
1-37 42 1645 0.33 6.7 46.6 0.2 90
1-38 40 1645 0.60 3.4 26.1 0.2 60
1-58 47 1650 0.38 24.0 40.9 0.2 36 Dark, medium grain size deposit.
1-59 47 1775 0.38 24.0 44,0 0.2 180 2890 4.1 450(65.2) 10-15 am crystallite size,
some large crystals.EDAX: 61.8w/o Si and 0.1 w/o Ctstrong(322) and (222) orientation.
-17-
A typical chemical vapor deposition system contains several variables
such as total pressure, partial pressures of various gases, gas temp-
erature, gas composition, substrate temperature etc. These parameters
are usually interdependent. In addition, in many systems such as
Si3N4 , there are reactions in the gas phase that are not fully under-
stood. Thus, in our system, it was usually difficult to control
these variables in a perfectly reproducible manner. These difficulties
probably resulted in the range of crystal morphologies described in
Table 2 for seemingly similar deposition conditions.
Secondly, the deposition rates, measured by determining the coating
thickness, were subject to considerable error especially when rough
deposits were obtained. Very often, the deposits did not adhere to
the substrate, and no measurements could be made. However, these
findings pointed out the need for a much better control of process
parameters, especially the gas composition, before any correlation
could be attempted. Experience gained in other programs also suggested
that in a furance deposition process, very often the furnace walls
would also be coated. This would then significantly affect the heat
transfer in the gas stream from one run to the next.
We were successful in making fine grained Si4 Nt. deposits as shown in
Table 2. An example of the crystal morphology is shown in Figure 4.
The size of the crystalliets varied widely, even within a given sample.
For example, the crystallite size in Figure 4a is between 2 and lOim,
while in Figure 4b, it varies from about 10 to ;O0w.m. Mechanical
-18-
I
A
A
FIGURE 4 Morphology of SiN,, madO, with nitrogen as the carrier gas.
A) 2000X
B) 200X
' # -i -19
property evaluation of some of the deposits gave reasonable results.
Hardness values ranged from 2700 to 2800 (HV2 90 ) and fracture toughness
from 2.6-4.6 MPafm. As shown in Table 2, the deposits were usually
oriented, with no particular orientation consistently dominant from
sample to sample. EDAX analysis for Si and Cl contents revealed that the
deposits were probably stoichiometric in silicon. For pure silicon nitride
the stoichiometric proportions of Si and N are 60 w/o and 40 w/o respectively.
Some samples showed the presence of small amount of chlorine. It is
not clear if the presence of chlorine is due to residual, unreacted
silicon halide, and whether chlorine is present at the crystallite
boundaries. Since EDAX cannot detect nitrogen or oxygen, it is not clear
if all the silicon is tied up with nitrogen or whether some SiO 2 may also
be present.
Attempts to correlate deposition rates with process parameters for specimens
made with argon as the diluent gas were also unsuccessful. The various
parameters are shown in Table 3. The pressure was maintained at 25-30 torr
and the velocity of gases was maintained, in one set, at 21-25 m/s. The
variable in this set was the SiCl4 /NH 3 ratio, with AGPP at 1.2 ± 0.1 torr
and hydrogen partial pressure at 6.5-7.5 torr. Again, there was no correlation
possible.
Microscopic evaluation of deposits in this set showed a wide variety of
deposit of morphologies - from highly oriented whiskerlike growth to
fine, equiaxed crystallites. Figure 5 shows an example of the range of
morphologies obtained. Most samples showed a crystal morphology similar
to that in Figure 5a. However, in some areas of the coating in a given
sample, a very fine grained deposit was obtained, as shown
-20-
1AMBLE 3 Summary of results on Si ,N,, made with argon a tile diluent gas
Gas TRSTotal Substrate A P Velocity SiCI. Deposition KV? 3 point
Run Pressure TSmp. GPP H2 at TP N113 rate kg/mm Kc PaS Tort. K Torr. Torr. MiS ratio -mhr (Load) MPa/m (kai) Remarks
1-2b 28 1525 1.10 7.3 22.1 2.9 204 Coarse crystallites.
1-27 25 1525 1.70 6.4 24.7 1.8 126 fine, opaque crystals.
1-28 25 1520 0.88 6.5 24.6 3.6 - Preferential growth of spikes due
to hetrogc lius nucleation on
the matrix of fine crystals.
1-30 28 1525 1.14 7.3 22.1 2.9 - Poor deposit, whisker growth,
white needles- appear amorphous.
1-31 28 1475 1.14 7.3 21.4 2.9 - Mixed amorphous type and
crystalline deposit.
1-32 28 1475 1.30 7.3 21.4 2.75 360 Clear, -oarse crystallites.
1-33 28 1475 1.30 7.3 21.4 2.75 156 Coarse, amber crystallite5which appear transparent.
1-3 28 1425 1.30 7.3 20.6 2.75 102 Clear amorphous deposit.
1-39 67 1650 0.55 3.1 45.5 0.2 126 3210 3.55 Depositon a vertical disc,(200) uniform, fine deposit. Analysis
(FDAX) show. 60.7 w/o Si,0.2 w/o C,
1-40 h7 1645 0.55 1.7 45.4 0.2 126 Mixed crystal size along the
periphery of disc. Range 10-12
.m-shows strong (002) orientationin XRD.
1-41 29 1645 1.35 7.5 86.2 2.75 300 as above
i-4 2 29 1545 1.35 7.5 81.0 275 -
-43 28 1525 0.80 4.4 35.9 2.75 186
1-44 28 1535 0.80 4.4 36.1 2.75 - Strong (222),(322) and (304)orientations. Coarse (2O,.m).rystals otid large plikos. Pooradhes on.
L-45 50 1n25 0.72 4.0 42.8 2.75 102 2930 Coarse (15-20i,m) crystalliteb(100) on fine matrix. Amorphous
white deposit on top. latrlxcrysta!s 0.5-Lm.
i -21-
41
AdA
I 44
a -4
00
in Figure 5b. The average crystal size in this photograph is less
than lum. In one sample, long whiskers of the type shown in Figure 5c
were obtained on support rods. The fine crystal facets on individual
needles suggest the possibility of a very fine grain size. Again, the
lack of correlation of process parameters with deposit characteristics
must be attributed to the difficulties in controlling the reactions in
the chamber.
It is of interest to examine the SiCI4/NH 3 system. The reaction
3SiCl4 + 4NH3 - Si3N4 + 12HCl (1)
shows a change in the standard free energy of formation of 9,136 cal/mol10
at 500K. Thus, the reaction, if it were allowed to occur, would
theoretically be complete at a very low temperature. In reality, however,
the tendency for the reactants is to form an intermediate product,
silicon di-imide, Si(NH) 2 . Nihara and Hiral 11 have suggested the
following sequence of events:
SiCI4 + 6NH 3- Si(NH) 2 + 4 NH4Cl
6{Si(NH)2)n 6115° 42{Si 3 (NH) 3 N2 }n + 2NH 3 (2)
925K 3{Si 2 (NH)N,)n + NH3 1475Y
2Si 3N4 + NH3
-23-
Another mechanism that has been suggested is 12
SiCl4 + NH3 - SiCl 3NH2 + HCl
SiCl3NH2 + SiC 4 Si2NHCI 6 + HCl
or
SiC1 3NH2 + NH3 -* Si(NH2cl) 2 + HCI
These intermediate species undergo further interactions by successive
collisions and form more complex intermediate molecules containing
increasing number of N atoms. In these reactions HC is believed to
be eliminated successively in the vapor phase before the gases reach
the substrate surface where formation of Si3N4 is believed to occur.
Lin 23 , in his mass-spectrometric investigation of the intermediates
in the SiCI4-NH3 system detected the presence of SiNH2 Cl2 ions
and several other ions.
In any case, whatever the mechanism of intermediate reactions, it is
clear that these events occurring in the vapor phase are difficult to
control since they depend on intermolecular collisions. Therefore we
decided to explore the possibility of using SiF 4 as a source of silicon.
While SiC14 and SiF 4 are very similar in chemical nature, their
reactivities are quite different. The most important ions derived
from the first comuination of two reactants are similar in SiCl4-NH 3
and SiF 4-NH3 systems,13 but the successive collision products are
quite different. The higher reactivity of the chloride appears to
-24-
accelerate the formation of intermediate polymeric molecules containing
several NH and NH2 groups while in the fluoride system ions containing
more than one NH2 groups are not observed, 14 Thus, it might be possible
to minimize the vapor phase reactions in the SiF4-NH3 system and achieve
a better control of the deposition process.
The results of experiments made with SiF4 precursor are shown in Table 4.
Again, we encountered problems in correlating deposition parameters
and deposition rates. Some indications were, however, obtained that
the rate increased with substrate temperature but decreased with an
increase in the AGPP, other conditions being identical. Very often,
the measurement of deposition rate was rendered difficult due to non-
uniform coating thickness along the length of the bar. This clearly
suggested the possibility of non-uniform temperature and gas composition
along the axis of the reactor. Another problem which could contribute
to this variation was deposition on the furnace walls.
The mechanical characterization of the deposits was more extensive than
in the previous sets. Hardness values ranged from about 2500 to
3150 (HV500) and the fracture toughness was 3.6-5.0 MPavm4 . The flexure
strength values were between about 80-210 MPa, obtained on bars tested
in an as-deposited conditions. Figure 6 shows a typical deposit made
with SiFi, precursor. In one case a crystalline deposit of an average
size between 5 and 10 microns is obtained (Figure 6a). The formation
of powdery surface layers is shown in Figure 6b. This particulate
matter appeared to be adherent to the matrix surface.
II -25-
S - C, V
- , 0 * A , 4
04 0 C M 0 4
0. C C,..0.,> 0> C. C 0
-c -* - 04 .~. 4t04 O O.
B
(b) Silicon deposition by disproportionation of a subhalide
The experiments described earlier were not successful in
making a CNTD deposit of Si3 N4. The nature of the CNTD process tends
to suggest that an intermediate, polymeric product might be required 5
to achieve the grain refinement. While SiCti, might readily lend itself
to such a reaction, its high reactivity to ammonia makes it difficult
to control the reaction near the substrate surface, rather than in
the gas stream.
Rochow 15 has shown that a reaction between SiC1 4 and H2 may lead
to the formation of an intermediate subhalide with an approximate
composition SiC12 .6 1. This precursor is believed to lead to the
formation of CNTD silicon carbide. '+'5 At this point it became
obvious to us that the conventional CVD system was probably not suitable
for the grain refinement of Si3N,,. We, therefore, sought alternative
approaches to the conventional one. One possibility was to deposit
elemental silicon and then attempt its nitridatlon.
While it may be possible to deposit elemental silicon by a variety
of methods using different precursors, it appeared to be appropriate
to use the same basic system of SiCI. -H2 or SiHCI3 -H,' that we had
extensively used in our investigations of silicon ceramics. Also,
there were potential advantages in depositing silicon by first forming
a subchloride and disproportionating the same over a substrate. The
most obvious potential benefit was grain refiliVmeut if the disproportion-
ation could be carried out simultaneously with nitridation.
-28-
We studied the feasibility of this approach for the following reactions:
SiCl4 (g) + Si(s) 2 SiC12(g) (4)
2SiHCl 3 (g) + Si(s) * 3S1C1 2 (g) + H2 (g) (5)
The lower halide is produced by passing SiC1. or SiHCl 3 over hot
semiconductor grade silicon chips in a graphite crucible. The variation
of the standard free energy change of the reactions with temperature
for the silicon chip crucible is shown in Figure 7.
The equilibrium constant of the reaction, K = exp (-AG/RT), can also
be expressed in terms of composition and total pressure, and, for the
two reactions considered above is given by:
K4 = SiCI2 P (6)n SiCli [ Si C12 + nSCl 4 + n]
SiC 2 . nH2 P 2 (7)K5 =-n----HCI + +
LSiHC14 SiCl 2 +nH 2 +nSiHCl 3 +
Where ni is the number of moles of the ith species, nI is the number
of moles of the inert (carrier) gas(es) and P is the total pressure.
Using these expressions, the equilibrium degree of completion of the
reactions was determined for different values of the parameters n I/nR
and P where n I/n R = the ratio of the moles of inert gas to the moles of
reactant gases, or the degree of dilution of the gases. The results are
plotted in Figure 8 and 9.
-29-
iool
807~
86&-
40-1
-~20-
05
0-
0U-4 H
3-40-
-fRef. #9
-1001 L 1 ~1000 1400 1800 2200
TEMPERATURE, K
1. 3Si(s) + 2N2(g) - Si3N4(s) Ref. #9
2. Si(s) +- N3g S13N,4(S) + 2H2(g)4 1
3. SiC14(g) + 4 NH3(g) - -1 Si3N4(S) + 4HC1(g)
4. 2SiHC13(g) + Si(s) - 3SiC12(g) + H2(g)
5. SiC14(g) + Si(s) -~ 2SiCl2(g)
4 16. SiF4(g) + - Nli3 (g) - 3 Si3N4(S) + 4HF(g)
7. SiCz.(g) + -1S13N4(S) - 2SiC12(g) + 2~ N2(g)
8. SiFi4(g) + H2(g) - SiF 2(g) + 2HF(g)
FIGURE 7 FREE ENERGY OF FORMATION AS A FUNCTION OF
TEMPERATURE FOR VARIOUS REACTIONS.
-30-
0000
N009T0
-44 C-4
-104 E4 0
+4- )009T' w
E-4
Cd~~C F '-4
C.' NOOE T41-
CC,
-31-
II I -T [ l l l
0.0
1 . 0 -... _ ._ ._ .. ,. .. - ........ .
Q. 0.95
//,/ - ni~nR , torr
//7 1 38! / 10 38
A, 1 380A20 380
/1 / 0 1 610iiII20 610
0.90 6 1 760 --44,-4 A 20 760
2 SiHC1 3 (g) + Si(s) i 3 SiC1 2 (g) + H 2 (g)
L!nSo 0 0 0 0
r4 -
0.85I0-I 1 10 102 103 104
FIGURE 9 EQUILIBRIUM DEGREE OF COMPLETION OF SiHC13 REDUCTION AS A FUNCTION
OF TEMPERATURE, PRESSURE AND DEGREE OF DILUTION.
-32-
Considering reaction (4), it is clear that a substantial amount of
SiCI 2 can be generated above 1500°K if the total pressure is reduced
to 40 torr and sufficient dilution of the gas stream is carried out
(Fi&ure 8.) Having achieved the formation of SiC12 in the Si reservoir,
the reaction can then be reversed near the substrate simply by drop-
ping the temperature to form a deposit of silicon, which may then be
carburized or nitrided as the case may be. In this approach, the
purpose is to cause a disproportionation of the lower chloride which is
believed to result in a finer grained deposit of silicon, than would
result from direct reduction of SiCL by hydrogen. Calculations
indicate that the yield of silicon by the latter reaction will not
be significant at 15000K and 40 torr with dilutions up to 20:1.
Similar considerations for reaction (5) involving trichlorosilane
show that the formation of SiCI 2 is very energetic over the
temperature range 1300°K - 17000 K, indicating that SiHCI 3 may
be a more efficient source of SiC1 2.
The use of a silicon reservoir imposes an upper limit of =16750K (M.P.
for silicon) for the reservoir temperature since the presence of
liquid Si would create handling problems in the reactor. This
problem may be circumvented by using Si3N chips instead as the
reducing agent, since it is more stable at these temperatures.
The reaction can then be written as:
-33-
1[2SiC1 2 (g) + 2(gSiCl 4 (g) + - Si 3N4 (s) -- N2 (g)
Figure 7 shows that reaction (8) is feasible only near 20000K
under standard conditions. Figure 8 reveals that this reaction will
proceed to completion at low pressure and high dilution. Once SiC12
is formed via generation in the chip pot, thermochemical data
suggest that it is relatively easy to cause disproportionation and
subsequent nitridation using a suitable source. The reaction for
the formation of Si3N4is:
3SiCI 2 (g) + 4NH 3 (g) - Si3N 4 (s) + 6HC1(g) + 3H2 (g) (9)
This reaction will proceed very energetically even at room temperature.
Alternatively,
2SiCl 2(g) + Si(s) + SiCl 4 (g) (10)
3Si(s) + 4NH3 (g) Si 3 N4 + 6H2 (g) (11)
3SiCl,(g) + 4NH3(g) : Si 3 N4 + 12HCI(g) (1)
Calculations of standard free energy changes show that rfaetiors -01)
and (1) will both occur with nearly equal ease in the range 1500°K-180 0 °K.
-34-
These reactions should also occur favorably with nitrogen instead of
ammonia as the nitriding specie although with much less vigor. Reactions
(11) and (1) may also be carried out at much lower temperatures; however,
one is then concerned with the rate of deposition and morphology of
the deposit.
Several runs were made in this study as shown in Table A-4, Appendix I.
Table 5 summarizes the results of evaluation of the deposits.
The deposition was carried out by passing SiCLf-bearing carrier gas
through the reservoir containing silicon source material (pure 91, or
scrap RBSN). Several experiments were made in which the SiC12 formed
in the reservoir was allowed to mix with ammonia near the substrate.
In other cases, SiC12 was merely allowed to disproportionate to form
silicon. In some experiments the reservoir was empty but was used
as a preheat chamber for SiC14 which subsequently reacted with NH3.
We made several runs at high chamber presures (450-600 torr) to assess
the effect on deposition rate. In these runs, the nitrogen flow rates
were varied from 0 to 10 liters/min. In those runs where no nitrogen
was used, we did not get any deposit although argon was used as the
diluent. Most of the deposits were powdery, loose and difficult to
evaluate. Figure 10a shows the deposit obtained when SiC14 was passed
through an empty silicon reservoir being used merely to heat the gas
to about 19000 K. The deposit has a crystalline morphology with a size
of about 10-15pm. The mechanical properties data suggest that the
-35-
TABLE 5 Results of tests on samples deposited in the study of silicon halide disproportionation
RUN kg/mm2
K TRS.MPaI (Load) MPa- (kei) Remarks
569 White powdery deposit573 1085 (50) Thick powdery gray deposit574 Grey silicon deposit under a layer of yellow
powdery layer.
575 600 124 (18) Fine crystalline Si regions of botryoidal morphology.
576 Powdery yellow deposit.
577 2645 (100) Non uniform deposit. Partly powdery yellow, partlycrystallLne L
_gstlii~ rey deposit.578 1035 (100) Grey, powdery deposit with metallic lustre.
579 3235 (100) Non uniform deposit containing yellow powdery andgrey coherent areas.
580 XRD shows riSi N (some). Non-faceted grey crystallinesurface.
581 1005 Non uniform nitridation, XkD OSL3N,, 4-SiC. Top3401 (500) of bar shows aSiNs. OSiC and Si.
582 XRD q-Sie, c-SiIN4, some .SiC. Dar metallicrovf surface with fine graj d rexrns._
583 A mixture of grey coherent botryoidal deposit and.__ an d y ell ow o wde ry deposit. ____________
585 XRD: a-Si1 3N., SSiN. Mostly powdery deposit. someneedle-like areas. Dissolved in Hf I HNO. - --
586 Coherent, fine grained deposit with some :onicaltopograhy o the top of bar. ____
587 Mixture of fine sod coars- globuler crystallites.columnar crystal habit.
588 XRD: uSi IN,, Slo. Bottom-shiny grev crystals, top,grey powdery de-osit. ___
589 XRD: Graphite, grey powdery deposit - XI(D shows
graphite and some SiN*.
590 XRD: -SiiN, + graphite. tdreenish white powderydeposit.
591 Greyis white powdery deposit.592 3380 (500) XRD: nSiN4 with traces of -SijNu. White soft,
fluffy deposit.
593 Mixed fine-grained and needle-like_dposit.
594 3200 (500) 138 (20) Mixture of coarse and fine faceted cryta~ittes.595 2930 (NC-350 4.15 138 (20) Coarse and fine faceted grains, good adhesion to
3320 (rarrett SN) RBSN substrats. _-----
596 2460 (500) Coarse and finegained crs'stallites.597 2990 (500) 5.8 244 (35.4) Fine grained depoi_.with ,Lor adhe.ion ,
598 Fine needle shaped crystals dispersed in a fine--__ powdery deposit.
599 Fine needle shaped crystals dispersed in a finepowdery deposit.
600 2710 (500) 4.4 143 (208) Dark, powde_ _surface layer ona coherent dark denosit.
601 3270 (500) 3.6 Dark faceted deposit on to resto st. ao TA-602 Loose whisker-like crysta;.jites on a cryst lline deqs.
603 Deposit similar to run I 598 and 599.604 Deposit similar to run I 598 and 599.
605 Mixture of fluffy whisker-like owthn.__ pyery denoos -606 Black crystalline deposit and white fluffy regions. Some
areas show "fused" spots.
"60 7 No deposit.
608 No deposit.
609 No deposit.
610 1150 (100 Matrix of fine grained deposit covered with yellowpowdery layer.
61l 1200 (50) Same as 610.
612 1075 (100) _-Same as 610.
613 Dendritic growth near top, coarse faceted crystallites
and fine rev deposit.614 Transparent loose needle-likeIdeosit.
615 ph whisker-like deposit.
616 .i...... . ..... ............--- -- --- --- ---.... ..... ... .e l .l . ..
,1i 21,90 (.Of)) 3.8 MiXL t. oi d.-like and ic tedo cry sta morpho ....618 Smooth, amorphous looking deposit with some fine
crystalline areas.
FIGURE 10 Morptio I(),, ,g ' Ipitni ill tt ilicon hal ide
disproport i Omm ti'm ,t od. Wit h s ic
A) 'Ofl/2")()OX
B) 600/ 3 ) /
material is CVD Si 3 N4 (See Table 5, run #597.) Figure lOb, on the
other hand shows the structure of sample #610 which was made by passing
SiC14 over CVD Si 3N4 chips but without any ammonia. The deposit
is powdery on a matrix of fine silicon layer. This and other samples
that we could test showed hardness values of HV 1075 to HV 1200, sug-
gesting that the material was probably silicon. Thus, despite the
thermodynamic feasibility of nitridation of silicon,(Figure 7) we did
not succeed in nitriding elemental silicon with nitrogen.
In other experiments where ammonia was used, we could form silicon
nitride. In some cases traces of 6-SiC were found in the XRD patterns.
In one run (#589, see Table 5) where no nitrogen source was used, the
deposit exhibited a B-Si 3N 4 pattern, indicating that the Si3NL4 used as
a silicon source had probably decomposed and redeposited.
In general, the results of disproportionation study were rather disap-
pointing. The three runs made with SiHCI 3 instead of SiC14 resulted
in a silicon nitride deposit which showed a CVD type crystal morphology.
Figure Ila shows the deposit of Si 3N, with a crystalline morphology.
Several large rounded and faceted crystallite are observed on the
matrix of platelets having an average dimension of 50-]0im. Another
sample, shown in Figure llb, appears to have a rounded, botryoidal,
morphology. However, it became obvious that deposition of fine grained
SijN via disproportionation of SiCl, and subsequent nitridation would
be very difficult.
-38-
r__ -7
Ilif
FIUE1-opooyo kpsbtnd nteslcnhld
Aipootin tinsuYw hSl~
A) 7O
B 0X/:o
(c) "Alloying' of SigN, for grain refinement
The idea of introducing a parallel reaction in the gas stream
to curtail the columnar growth of a deposit during chemical vapor depo-
sition is not new. Researchers have effectively used this technique
for example, for the grain refinement of tungsten. 17 '1l In our
experiments we decided to use an essentially similar approach.
During the first year on this program 1 we had examined the feasibility
of using propane for this purpose. In that study, we attempted to
co-deposit SiC and Si3N4 on the resistivety heated tungsten filaments.
Although we could not succeed in doing so, we found that frequently
the presence of propane in the gas stream probably led to a relatively
finer, columnar crystallites. We, therefore, used this approach in
an attempt to reduce the crystallite size of Si3N4 made in a furnace.
The run conditions for these experiments are given in Appendix I,
Table A-5. Table 6 summarizes the results of evaluation of deposits
made with propane additions. Again, no specific trend can be detected
as far as relating deposition rate with process paramenters. For example,
Runs #51 and 55 were made under identical conditions but the deposition
rates were different. XRD showed random orientation of deposit in
one case while the other showed a stronly oriented deposit. Tho
photomicrographs (Figure 12) showed typical, well-faceted deposits
Fi&ure 12a shows the surface topography of sample #50 (Table 6). The
crystallite size is 5-10m. By comparison, sample #56 (Figure 12b)
shows massive crystals, about 30um in size. Table 6 shows that for
sample #56 the active gas concentration (SIC]i, + NH-) is much higher
than for sample #50, and the deposition rate is more than doubled.
-40-
A
B
I FIGURE 12 MorphiolIogy of Si N vniIdt wi th adi(Ition ofl f propane.
to tit he a tlr I
A) I ()()(X / O )UX
1j) I ()t)()x
I-r :
= - ' - .-.
C . . S ..
.. . . .. ..c P. tu. . .n, Zr
a
summarized in Table 7. The samples generally showed poor adhesion to
the substrate, and strong crystal orientation, in XRD. We did not
find any trace of SiC, or carbon in the deposit. The crystallite size
varied considerably, as shown in Figure 13. Comparing Figures 13a and
13b an increase in t,.e active gas concentration may have resulted in
an apparent grain refinement to some extent. However, these experiments
also proved to be unsuccessful in producing fine grained Si4N4 with any
consistency.
We, therefore, decided to terminate these efforts, and attempt co-dep-
osition of AIN. Aluminum nitride is deposited by causing a reaction
between AlCI3 and NH,:
AICI(g) + NH (g) -AIN(s) + 3HCI(g) (14)
The study of this system was being carried out on another program
sponsored by AFOSR.7 We were conducting experiments to refine the
grain structure of CVD AlN by introducing silicon in the gas stream.
A natural extension of this work was to explore the feasibility of this
approach in the system SiNu + Al. Some of these experiments were
carried out under the AFOSR program. Since this work was directly
applicable to the present program, the details of these experiments
are included in this report. Table 8 gives the summary of results
obtained from the samples in this work.
The sample in run #652 was made without any additions of Al and resulted
in a typical CVl) Si Nj, deposit. Two samples from this group were
-43-
BA
FIGURE 13 Morplogy,,v, ') \md r withIio~I n
16 andl nmnon i ,i
(A) A(; IT' ).,I',t (B) A(J 1 1.7 torr
The higher concentration of C3H8 apparently only introduces greater
amount of hydrogen since no carbon or SiC was detected in the XRD
pattern for sample #56. The interrelationships of the various parameters
in this case are interesting but unknown. Thus, in general, the addition
of propane did not appear to have much beneticial effect on the grain
size of the deposit.
We made another attempt to introduce a carbon source to influence the
reaction between Si and N2. Methyltrichlorosilane is an excellent
precursor for the deposition of SiC. 4 The reaction is:
CH.jSiC1 3(g) SiC(s) + 3HCI(g) (12)
We attempted to influence this reaction with the addition of ammonia.
We were interested in the following possible reaction, although other
reactions sequences were probably also feasible.
CH3SiCI3 (g) + A NH3(g)- lSI3N4(s) + HCl(g) + C(s) + 2H2 (g) (13)
Alternatively, it might be possible to combine reactions (12) and
(13) so that co-deposition of SiC and Si3N4 could occur. Calculations
of the change in free energy of the reactions suggested that the
combined reaction would be more favorable than reaction (13) under
standard conditions. If this could be achieved under reduced pressure,
and other parametric constraints, we might be able to utilize the
differences in the reaction rates of formation of SiC and Si-N, to
achieve grain refinement. The results of these experiments are
-45-
IC I
I C -CC~ I.C C
I *~1 2 r 0 52: a-
C' t-j 0
-- a'-"Ic 'C4
ICCA ?~0 I
C' 4 CC
CC CO -Va;C'' I
22-2
'-C YCCI C I CC ' r CC,~a I ~ Zo
7 00 ICr. I
C' Fi 'a C.-'
2li ~- I I.0 N
3C , I
'0 0d
a'- oQ'C~ a; -, a a;CCC' 0 (1 -a a;- C' A - a; a; a;
C a; '"C. I-'C I -~'C CaI..C~ - C'~CC' -.- , - a; -a -CC (3--.-'S -7 -aC' CC
I C
-. -to' - I C' C'
o C..-'
0' ~I'~- a; -~ ' 4 -CC ~
C'a . - -
C. C - -C.C. 12: FC ~ C C
.0 '~ 0 '0a - - -
4 C F
ICC a o a aoat- .7C-'C' I
.4 - ~- C' C'C' 'C'0 ' -a -a*0 § p I
1-' C' - ' C I-46-
1)
[1
TABLE 8 Summary of results on deposits made with Al as "alloying" addition
Substrate Deposition HIV TRS
Run Pressure Temp. Rate kg/rn" K MPa0 Torr
0K wm/hr (Load) MPazm (kai) REMARKS
8-652 65 1600 100 3255 - 376 Non-uniform deposit morphology from(100) (55) top to bottom, from fine crvstallites
to coarse faceted crystals.
8-653 65 1625 2960 - 158 same as above.(200) (23)
8-656 65 1595 - - 290 Glassy deposit with rounded domes
(42) and areas of faceted crystallites.Poor adhesion and integrity.
8-655 65 1590 3410 - Fine grained deposit with rounded(100) domes on the surface.
8-656 65 1590 2560 - 129 Amorphous type deposit with microcracks.(500) (19)
8-657 65 1585 250 2330 4.9 - Translucent deposit with fine grained
(500) domes on the surface.
8-658 90 1595 2320 3.2 - Dark shiny deposit with domes.(200) Fracture surface appears glassy, non-
columnar.
8-659 90 1590 Fine-grained. columnar deposit,
with surface cracks.
8-660 50 1420 Whitish amber deposit on top, dark
at the bottom, appears glassy in
fracture.
8-661 62 1425 same as above.
16-182 73 1750 74 Fine-grained deposit, cracked near
top on graphite substrates.
16-183 73 1750 - Whisker-like growth on the entire
surface.
16-184 72 1750 146 Fine grained, uniform deposit.
16-185 75 1750 174 1945 3.2 204 Botryoidal morphology of deposit,
(29.6) cracks on the surface. Size of rounded
crystallites varies widely (iO-50um).Some porosity in the deposit.
16-186 75 1735
16-187 76 1750 150 Lavered deposit showing a mixture of
dense columnar growth, porous banded
region and preferential growth ofhexagonal platelets on the surface.
Generally columnar, coarse rounded
crystallites. (20-50m)
16-188 76 1750 142
16-189 74 1750 105
16-190 72 1750 - 2170 5.3 225 Faceted crystalline deposit surface
(300) (32.6) with columnar grains. Surface shows
randomly oriented platelets with
edges nearly normal to the surface
of deposit.
i1 16-191 76 1750 194 1890 3.5 168 Finely cracked, botryoidal deposit
(24.4) with size of rounded crvstallites ranging
from 10 to 301m.
16-192 79 1710 149
16-193 91 1750 182 Layered deposit showing columnar,
porous, layered and granular
morphologien in the fracture
cross section.
1!
: -47-
analysed extensively to study the effect of incorporation of Al.
Fi&ure 14a shows a typical CVD Si3N 4 deposit with columnar grain
structure. In comparison, the grain refinement achieved by the
incorporation of Al in the material is clearly visible in Figure 14b.
The distribution of Si and Al in the material is shown in the X-ray
elemental density maps (Figure 15). The complimentary variation of
the concentration of Al and Si is clearly visible. This suggests
that both Al and Si were incorporated simultaneously since the
<4 elemental map for nitrogen showed a uniform distribution in the section.
This result clearly indicated the possibility of refining the grain
structure of Si3N4 by adding Al to the system.
Further work was carried out to select optimum deposition conditions
for the incorporation of Al (Runs #182-193, Table A-7, Appendix I).
In any CVD operation the compatibility of the coefficients of
thermal expansion of the substrate and the coating is an important
consideration. When AlN is incorporated in Si3N4, the value of the
coefficient of thermal expansion, a, of this dual-phase coating is
different from either of the constituents. This presents some
problems in maintaining the integrity of the coating. We addressed
this point by studying the nature of deposits on various graphite
substrates and hot pressed silicon nitride bend bars.
The microscopic evaluation of the coatings revealed that although some
grain refinement could be discerned, there were problems related to
the deposition that were difficult to control. For example, as shown
-48-
A
FIGURE 14 Morphology of SijN, made (A) Without and (B) with Al addition
Magnification 200X
------- --
0o
ai
-4
Ez..
in Figure 16a, we obtained a deposit which was practically free of
distinct columnar grain structure although the surface revealed the
uneven botryoidal topography (Figure 16b). On the other hand, another
sample showed a considerable variation in the fracture topography
(Figure 16c). The initial deposit was columnar, but became progressively
porous. The top layer (about 10pm thick) almost totally delaminated
from the rest of the coating. This layer also had an unusual crystallite
orientation as revealed in Figure 16d. These variations in the
morphology through the coating suggested non-uniformity of the deposition
conditions during the run.
These experiments were successful to the extent that we could demonstrate
the possibility of CNTD-type grain refinement in Si3N4 by incorporating
a suitable second phase. Detailed evaluation of these samples and
further exploration was not carried out due to the constraints of time
and funds. However, there is no doubt that a more thorough exploration
and understanding of these reactions in a systematic manner is warranted
and should be continued.
(d.) Deposition of Si3N4 on bend bar specimens
We made a series of runs in which conventional CVD silicon
nitride was made and deposited on HLM graphite and RBSN substrates.
The purpose of these samples was to evaluate the CVD silicon nitride
made in an indirectly heated furnace and compare the results with
those obtained on tungsten filaments.1 We tested some samples for
strength, hardness, fracture toughness and crystal morphology. Some
-51-
I:I
rg
I
FIGURE 16 Morpho i ogv of Si N dtip)s ite d in the alt minum dopant study.
(A)
C-- - .. .
'
FIGURE 16 Morphology of Si dN d 'pOsitLd in thc aluminim dopant study.
(C) 1000X
(I)) lO5OX
-53
b'i''~ . . . .. .. I I - " I I .. . 1 7 I I
samples were sent to Mr. R. Rice of Naval Research Laboratory for eva]-
uation. The results of our tests are summarized in Table 9.
Figure 17 shows the morphology of a typical deposit in this set.
Usually, the attempts to obtain fine-grained deposits on resistance-
heated filaments resulted in amorphous or glassy morphologies. The
mechanical properties of samples in the present work were comparable to1
those in the earlier work.
The fracture energy tests carried out at Naval Research Laboratory
showed 19 that the CVD Si 3N4 made at San Fernando Laboratories
exhibited a fracture energy of about 20 J/m2 , comparable to values
obtained on materials from other sources. The calculations of fracture
energy on the basis of indentation fracture toughness measurements that
we carried out gave values in the range 17 to 86 J/m 2 with an average
of about 40 J/m2 . The difference probably reflects the relative sample
size in the two types of tests. There is another possibility for
the difference in the fracture energy values. The samples for
indentation fracture toughness measurement are polished ceramographically.
This procedure usually introduces a residual compressive stress on the
surface. We did not anneal the samples after polishing to remove any
possible residual stress. In addition, the error in the measurement of
crack length using the microhardness tester at 400X probably resulted
in an over-estimation of the fracture toughness by this technique of
about 15%. Rice Ili also noted that the fracture strength of the
samples was only about 69 MPa (10 ksi) and this was related to the
large grain size of the deposit. Although we did not carry out a
-54-
TABLE 9
Summary of results on bend bar specimens
Hardness Fracture TRSRun kg/mm2 toughness (3 point)
# (load) Kc, ramr MPa (ksi) REMARKS
168 2505 (200) 6.1 392 (57) HLM graphite substrate
170 2215 (500) 3.55 123 (18) RBSN substrate
173 4130 (100) - 197 (29) HI1 graphite substrate
175 3480 (300) 165 (24) HLM graphite substrate
1
-- II
1 -55
-0J
44K
'4441" V" 'Y"%j4
FIGURE 17 Morphology of SiN 4 deposit on bend bar specimens.
(A) 2000X
(B) 2000X
-56-
dk-tii led inves ti gat ion of the rel-t ionshi p be tween strength and the
sizt- of crystallites, our results generally support his findings.
(e) Measurements of electrical properties
As discussed in the experimental procedure (p. 3 ),this was
not a part of work scope of the pre-sent program. However, the results
art- included here since there is an interest in Si N,, as a dit-lectric
*1 material. The samples for these measurements were made under another
program. The evaluation was carried out elsewhere. Table 10 gives
daita for our material. We also include data for other silicon nitrides
a.ad a SiAlON for comparison.
-57
IV. SUMMARY AND CONCLUSIONS
The objective of the present program was to evaluate various
methods of refining the grain structure of CVD Si3N4 deposited
on an indirectly heated substrate. We examined various deposition
chemistries to influence the reaction between silicon and nitrogen.
Experiments were conducted to study the effects of nitrogen, argon,
SiF4 , propane, MTS and aluminum on the process parameters and
morphology of the deposits. We also studied the possibility of
depositing elemental silicon by the disproportionation of silicon
halides and its subsequent nitridation.
We found that in most chemical systems studied the tendency of
SIC1 4 (or SiF 4 ) to form intermediate species in the vapor phase
by reacting with NH3 caused difficulties in controlling the process
parameters. This reaction resulted in deposits having a variety of
crystal morphologies and properties. However, the simultaneous
presence of aluminum chloride and its reaction with ammonia resulted
in a non-columnar deposit of Si3N4-AlN. Thus, although the attempts
to apply the CNTD process for depositing non-columnar, fine-grained
Si 3N4 were largely unsuccessful, valuable understanding was gained
regarding the chemistry of SI3N4 deposition. The salient points were:
(A) The silicon halide-ammonia system is not amenable
to CNTD-type grain refinement due to the propensity for vapor phase
reaction and formation of non-volatile intermediates.
(B) Additions of carbon, either as MTS or as propane, do
not cause grain refinement in Si3Ni4 under the experimental conditions
used in the present work.
~-58-
(C) The codeposition of A1N with Si3N4 apparently
permits grain refinement. This is probably due to the competitive
nature of nitridation reactions for Al and Si. A non-columnar deposit
of AIN-Si 3N4 is obtained.
The mechanical properties of Si3N4 made by the indirect heating of
substrates were comparable to those obtained on directly heated substrates.
The flexure strength values were 205 MPa (29.8 ksi) ± 80 MPa (11.6 ksi),
with values as high as 450 MPa (65 ksi). The hardness of the deposits
was usually in the range HV 2500 to HV 3000, with values as high as
HV 4130 and as low as HV 1900. The indentation fracture toughness was
about 4.0 MPa/m, with values as high as 5.8 MPaVrm.
In conclusion, the present work clearly demonstrated the potential of
vapor deposited silicon nitride in terms of achievable properties.
The concept of codeposition of a second phase to minimize or eliminate
the columnar growth habit was shown to be feasible in the initial
experiments.
An advantage of the vapor deposition technique is the possibility of
studying very pure alloy systems. Conventional powder technology
is very often limited by the presence of impurities, intentional or
otherwise. Therefore, it should be noted that the vapor deposition
technology offers the possibility of studying Si-Al-N system without
Ai having to deal with oxygen. The initial success of the codeposition
1, work warrants further development work to refine the system and study
the feasibility of depositing a range of SiIN,,-AIN compositions for
high temperature applications.
-59-
0 C C
x x
E-~ Lt~ r'
144
0
C14
00
P.4 4
po Cz ~-o C
60-
REFERENCES
1. Holzl, R.A.:"Investigation of the CNTD Mechanism and Its Effecton Microstructure and Properties of Silicon nitride," Sunmmarv
Report, U.S. Naval Air Systems Connand Contract #N00019-77-C-0395.July, 1979.
2. Evans, A.G. and Charles, E.A.: J. Amer. Ceram. Soc., 57, 371(1976).
3. Holzl, R.A.:"Grain Refinement by Thermochemical Means," Proceedings
of 6th International Conference on Chemical Vapor Deposition, Atlanta,Georgia, The Electrochemical Society, Princeton, N.J., (1977),p. 107.
4. HoLzl, R.A.:"An Investigation of the CNTD Mechanism and Its Effect onMicrostrurtilral Properties," U.S. Air Force Office of ScientificResearch Contract #F49620-77-0086, March, 1977.
5. Stiglich, J.J., Bhat, D.C., and Holzl, R.A.: Ceramurgia International,6 (1), 3 (1980).
6. Dutta, S., Rice, R.W., Graham, H.C. and Mendiratta, M.G.: "Character-ization and Properties of Controlled Nucleation Thermochemical Deposited
(CNTD) Silicon Carbide," NASA Tech. Memo 79277, paper presented at 80th
Annual Meeting of American Ceramic Society, Detroit, Michigan, May, 1978.
7. Panos, R.M. and Bhat, D.G.: "An Investigation of the CNTD Mechanism and Its
Effect on Microstructural Properties." Interim Report, U.S. Air ForceOffice of Scientific Research, Contract #F49620-79-C-0041, March, 1980.
8. Gazzara, C.P. and Reed, D.: "A Computed X-ray Diffraction Powder Patternfor Alpha and Beta Silicon Nitride," AMMRC TN 75-4, Army Materials andMechanics Research Center, Watertown, Mass., April 1975.
9. Kijima, K., Setaka, N. and Tanaka, H.: J. Cryst. Growth, 24/25, 183(1974).
10. JANAF Thermochemical Tables, Second Edition June 1971.
11. Nihara, K. and Hirai, T.: J. Mater. Si., 12, 1243 (1977)
12. Wannagat, U.: "Advances in Inorganic Chemistry and Radiochemistry,Vol. 6, p. 225, Academic Press, New York (1964)
13. Lin, S. - S.: J. Electrochem. Soc., 125, 1877 (1978'.
14. Lin, S. - S.: J. Electrochem. Soc., 124, 1954 (1977).
15. Rochow, E. (;.: "The Chemistry of Silicon," Pergamon Texts in InorganicChemistry, vol. 9, p. 1340, Pergamon Press, New York (1973).
16. HolzI, R.A.: Unpublished work.
I -61 -
17. Seymore, W.C. and Byrne, J.G.: "The Influence of MoCi! on the
Chemical Vapor Deposition of Tungsten from WFb," Proceedings of5th International Conference on Chemical Vapor Deposition, England,p. 815 (1975)
18. Landingham, R.L. and Austin J.H.: J. Less Common Metals, 18229 (1969)
19. Rice, R.: Private Communication.
20. Thorp, J.S. and Sharif, R.I., J. Mater. Sci., 12, 2274 (1977)
21. Walton, J.D.: Amer. Ceram. Soc. Bull., 53, 255 (1974)
22. Popov, E.P.: "Introduction to Mechanics of Solids," p. 202, Prentice-Hall, Inc. Englewood Cliffs, N.J. (1968).
1
4t .+ , ++
APPENDIX 1: Deposition Conditions
for S13N4
-63-
TABLE A-I Deposition conditions for Si3N,?, made with nitrogenas carrier gas for the precursors
Total Substrate Run DepositionRun Gas Flow Rate, ml/min Pressure Temp. Time rate0 SICx. NH, H2 Nj Torr. OK min om/hr
1-1 30 150 400 2700 25 1635 30 781-2 30 150 400 2700 25 1575 30 78
1-3 30 150 400 2700 25 1500 30 7
1-4 30 150 400 2000 25 1500 30 -
1-5 30 150 400 4300 26 1650 30 48
1-6 30 150 400 8600 39 1650 30 48
1-7 30 150 400 11,550 52 1650 30 48
1-8 30 150 400 11,750 52 1650 120 66
1-9 30 150 400 11.750 48 1620 60 48
1-90 30 150 400 15,000 64 1650 60 36
1-11 30 150 400 15,000 60 1675 60 18
1-12 30 150 400 15,000 66 1735 90 60
1-13 30 150 400 2700 25 1635 180 66
1-14 30 150 400 8600 40 1650 180 102
1-15 30 150 400 11,550 52 1700 30 96
1-16 30 150 400 11,550 55 1650 30 24
1-17 30 150 400 11,550 62 1650 45 54
1-18 30 150 400 11,550 6 1700 45 36
1-19 30 150 400 11,550 58 1750 60 72
1-20 30 150 400 11,550 58 1800 60 -
1-21 30 150 400 6,550 59 1650 30
1-22 30 150 400 8600 45 1650 60 6
1-23 30 150 400 8600 47 1650 60 0
1-24 60 300 15,865 78 1650 30 ISO
1-25 37.5 187.5 500 8600 48 1650 60
1-29 250 87 975 2400 28 1525 120 8
1-35 30 150 400 8200 41 1645 90 48
1-36 30 150 400 8200 40 1645 90 9
1-37 30 150 1400 7200 42 1645 60 90
1-38 30 150 400 4100 40 1645 90 60
1-58 30 140 4400 3720 47 1650 60 36
1-59 30 150 4400 3720 47 1775 60 10
-64-
04 $4 . 0 cli V) C14 0 1 00
0. ~ .C.4 - C1 ,- -1 r-4 -4 C) ,-4
.4 0 0 0 0D 0 0 0 0 0 C0 0 0 0 0)co9 - l C'J C14 '0 m cn 0' ON 0'. ' 0 c n ens m~ ej 'o
bo - r-i-i -4
V)4- L u' L, U'. in LI, LI, 0 L Lf LI, iI LI, .ICU.cJ q " C4 r- r- r- ej Lt) -T I? 'T N m "
V), Lr, LI, VI, 4z 4z \~.4 0 '0 '.D V) LI, Ln I LI00 'A .- 4 ,4 - - -4 -4 ,-4 -4 -4 1-4 14 -
0 $
44 Scu cUO I cc Ll, zI co OD OD OD 00 r- r- ON. 0% 00 Do4 m C14 C'! C14 C4' C"! 04 C'! C1! '.0 '0 C! C1! C'! C- LI,
0 4)E-- $
0c0c 0 0 0 0 0 LI, II 0 0 0D 0 000 0D 0 0 0 0D 0 %.0 '.0 0 0 0 0 (0
s1d -1? 'T-s -It -It -1 IT co 0 11 11 - . 00 00 co-q c'! C! ' c-! C4I C' N - C"! *q C- e-) C 4
$4
ON as N0 0 -~ 0 0% 0% O'. cc cc m% ON 0% 0 ON
0
t, 0n 0n 0 0 0D 0 0 0D 0
00L 0 cc 00 0 00 0 0 0 0 0I'o'4~ -4 en cn -4 1-4 ~ 4 -4 1-4
oULnI n U L n nL n L n W
'0 r" c0 0 1-4 en! IT . 0% 0 -4 C-I '4 - LA,N' C14 C') m ~ m' en- Cn M ~ en4 - ~
F4 -4--4 14 - -4 -4 -.4 -4 -4 -4 1-4 "A -4 -4
-65-
0
-4'
-4 -j C
M -- V) N7 00 N0 -- (7 Ln 00 I
0 10 .-4 00C C14 m 00 CC .,r
Z 4-4 0 C:) '0 '0 C:) (Z0 '0 C'0 C:)
Z
(a. t- r-- r- r-- r- r.,Wo - - - - - '1 - - 4 -4 -
z
G) I. C V Lf4 Lrn bfn 0 irn C C 'n C 'Q Its LIn CD N -T r- r- r-I C 4n *.r aN M
I- woj 'D0 'D0 - r- Lri Lr 'D r- Lr- 'D0 '. 'D$.
U, U)
4.r4 rim 5 CC a00CaCD 00CCAj ril 0 -T~ -7 -T I.T '0 m -:r It 00 m E-
CA
~ "4 0 0 0 M-M M-4 -4 -4 -4 -A -4 -4 -A r-4 LI1 Vf4 LPn LrEnU a% ON m' aN m' m as a'N z0 z. 'z Z 41
Z 04 N 4 N 1 N N CN N N- C. '.0 Z. .0
0 C.)
"-4 X
00'-4 r 4--
0.0
0 0
-4 -4- co
v- 4-U 0 0 0 0 0 0 0 C 4, -4 -4 . C)
0 0,
0 -0 004
"-4
o
-4 C/U fq -4 C-4 0 0 0 '0 0 0 A
00~~~U ON00) 0a
zU x
-4 -4 -A -4 -4 -q ,4 - H rI- 4 - K -
-66-
TABLE: A-4 Deposition condttions for the study of dtsproportlonatlon of silicon halide
Si
Total Reser voi r Subst rate Run Shtr~RUN Gas Fl, Rate, ml/rn Pressure Temp. Temp. Tine M.terlal.
# SIC;_ NIt, NZ _ _ Ar iillC e.___orr OK vr _d used
8-569 Soo 2 0 100 0 41 1975 1W25 10 UT- g' zraphi t9-573 275 O 0 0 2000 0 279 1950 I1115 11 VT- 2 2 tir pht, e
8-574 200 10 0 0 2000 0 290 1925 1300 30 UT-i2 grapiite
8-575 200 II0 125 0 2000 0 53 1975 118 30 L rT-2 ')t t
8-576 200 200 125 0 2000 0 58 1975 1273, P, 5 - 'aphl !
8-577 210 200 125 0 2000 0 60 1975 1425 (0 1- apI I t t,"
8-578 200 O 125 2000 0 0 60 1975 I - grapht I '
8-589 100 200 125 2000 0 0 5 1975 1435 4 ' -
8-'8C9 I,)0 2"() U 2200 0 0 55 1995 1R, .' 2I ,-ral t,.
8-"81 20 510 250 2200 0 0 57 1975 17 250-22 rt .]hlt
9-982 100 1600 0 2200 0 0 50 i975 L'T -22 y r apt, II
8-589 500 1 1) 2200 0 55 1675 1475 .10 -21 rahlte_
8-58S WOO 00 150 2000 0 0 42 1975 152 Pl 4'-'2 ;,. ap t t
8-58] 1O0 0 0 2200 0 0 37 1995 182, t',
1T-.2 g tC p oI
8-5A 7 10 5(50 250 2200 0 0 45 1 775 172S • '"-21 ral I,
8-58 200. 0 ( 2200 0 0 44 1975 16() 3, 11-22 gral'i t
•-,89 500 110 0 50 1925 112 IC- 22 - 6.erl ,.
"-"9 1OO 0 0 n 1200 0 40 1975 15', 60 LO- 7 r.- t',N,
-- 591 100 0 0 2200 0 0 45 1900 162'. 20 T-'2 , ap, i r
R-,92 IO0 510 0 2200 0 0 45 -,O0 161 ( 1'T-22. a, ih PON
P-191 IO0 5'0 J ) 2200 0 0 47 1500 1821 0 Vc- I r*'hI'
9-594 100 500 ( 2200 0 0 45 1810 IS NC- 3% S I rIC
8- 5 1300 Soo o 2200 0 0 43 1 180 60 NC-1 ,0 ,AI.F.ai'l l '"9~ -119F) Io 500 0 2200 o o 45 17 7', 30 NC:-,. I'- I t, .... 1,. ,IM,'-.,- 597 I90o 5(30 £) 2200 0 0 4S5 17 80 V, tl-2Naa FF %
-5198 100 5110 0 2200 0 0 46 1925 1810 W NC,- 3-.. rji.,IIC
8-99 I00 500 ( 2200 0 0 45 1910 181C is NC- 350, grav hit-'
8- " 100 2100 2 2200 0 0 S8 190 181 30 NC- I),. Riaph.IC
9-,101 100 1 20 400 0 584 1900 1800 t) NC- 3,:. rTC 1,"
8-605 300 1500 0 2200 0 0 50 1925 182,. 30 - 3.P, J ,T -
A-606 3 00 200 400 0 0 584 1900 1700 10 NG- 3%0, 25!ral;t,
8-604 00 0 -4 0 400 1 0 5 4 1925 1700 (40 NC-350 ,: iA- 'V 0 I ()I O) 600 400 0 O 590 1875 16". 30 ,p t
8-696 200 O 1200 400 1 0 597 192 170 215 Rap l
R6-5)9 50O U 1400 0 1000 0 40 1675 40 iraphLt'
851[1 500 ) 0, 0o 0 1 000 0 483 1675 40 1r lt,IIC
8-61 Soo 0 10,000 1500) 0 0 483 - 1700 I0 Arar,'lte
A-612 500 o 0000 5000 O 0 481 1710 30 rh te
8-563 5o 100 I0.000 53Oo 0 ( 483 171n 30
8-NI'., 500 0 200] 10,000 0 0 483 1875 30 gYapbU1
' , 'drod
9-61' S(1O 0 10,000 9000 0 0 457 1695 187 lT-22, N N0 I A"591
8'7 mla. t;I d.v- I Ir"3 hr. ntrl)d-tF v4
16000 . , . ,,w
9-Ni 0 0-1000 0 s5000 O 500 50 1550 1625 35 'T-22 Rr,-phlte
9-617 0 2)0 ' n 5O10 0 600 4(0 1525 1655 4' IT-22 Rraphlt
8-618 0 200 0 7)00 It 500 50 1475 I,50-1525 20 ITT- 22 Eraphl te
NOTES Il) The firt 12 runs tip to 8-583 were cond.cted under 1FOSR gr ta within the scope of a %imilar InvestIRatlIonin the SIC dep,Iltion.
(2) In all runt eOx',pt the liat I (i.e. 61its, 681. ), 13N1 %trap wast oed in the reservoir.
67
Oc.- P4j 93 00 o - IT CD
0 0D CDt -.* 0~ r- 0D 'D '
U)W -i -4 4-4 -4 1 - -4
cz W
w ~ ~ " 0 'i 0 '0 C0 '0 0 CD '0 r-0. CE-4 . , , ,
CC
C~~ 0 -. ' 0 0 0D 0Dm C l) c -Tj c, n m' w a'
(7 TO 11 C14
4-o 04 C 0 0) 0l 01 0 0 0) 0W- C ~-s~j CN -4 -.5 T
'4-. 0
-4
n Cn ,-4 111 Lfn Ltl Lf-4 -44 -4
0
. 4
4-)
I
0D -4 N1 C . L~' r, 0Lf.r Lfn L~ f Lr Lf-) U.U- LI') 'D
V I I I I I I I I I-4 -4 -4 - 4 4 -- 1 1-4 1-4
0
-4
41
E-400 1- 4 1-
U)
0
0)
CO J
E-44 ,4 -
-q
-l 0
J-O0 0 0 0 0
.1 ))D C
0 0400 0) C)
4~- -N '.
'4
'-4 ~ 010
0 E
CD C) NDc)cD) I -n c 0n
C) C) C-) C) C0 (D C) ) D C) CD'e, m- m Lt n n C) mn tn tn mn tn
(:0. 0 L-) C) -n c) c) u) 0n Ln C C) -nf" U) u') c) cfl c) ri Ifo ILr tn Lfl ") o
-1 ~CA-J C' C. N N ND G)N N
41
0 )
-H C)
r-( c) c c D0 r n c )
C)) (Nc) r U ) 'CP CN ) C) -)0) 0C4 > N N: C) C) C) C: NC
C)
C ~ ~ 0 0) C ) C ) ( C). C) C) CD
4- 0 CD4 0 C, 0 C) C ( C) C) m)
U) I - -A N I I N (N4 (N N (N (N (N0
U)N
HC) C) C) CD (D ) C) C) C) C) C) C)
o C C) ) C) C) C C) CD CD (D (D C)
co m m m co CO c c-i mX m m m m
I -4
'7 '
r r -f
:-C1 ar.10: a a
CDI
:k- Or 0c o- c 'J cc c c
C, i_ c _ __ __ __ _
APPENDIX II: Calculation of transverse rupture strength of a coated beam
The calculation of the transverse rupture strength of a simple beam is
accomplished by means of the flexure formula. When the beam is
composed of different materials, having different elastic moduli, the
calculations can be made by mathematically converting the composite
beam to that made of any of the constituent materials comprising the
original beam. Thus, when it is of interest to determine the strength
of the coating of material A on a bar of material B, the composite beam
is converted to a beam of material A by the method of equivalent
sections!22) In this method the cross section of the substrate (B) is
replaced by an equivalent section of the coating such that at a given
magnitude of axial strain, the forces developed in the substrate and
the equivalent section of the coating are equal. Then the entire
section can be treated as a single homogeneous material. The
equivalent section is generated by changing the dimension of the substrate
in the direction parallel to the neutral plane.
For a given axial strain e, the force developed on the substrate is
F = e.E .A . At the same point in the coating, the force is
F = e.E .A . For F = F , we have E .A = .A = nE A wherec C c s c c c S s c s
n = E /E • Thus the area of cross section of the coating to replaceS c
an equivalent area of the substrate is n.A s
For a bend bar of rectangular cross section as shown in Fiure A-I
the area of the substrate (A, = b. h.,) is transformed into an equivalent
area of the coating bv changing b, to n.b . Thton, the moment of inertia
-71-
1b
v-MATERIAL A
b2b
MATERIAL B
MATERIAL A
2 2*1 #1 - -(n .b , 2 t~
h 2 2
FIU'RE A-] METHOD~JI OF EOUI VALENT SEiCT ION
-72-
P
KA L/_
p/2 P,2
(a) 3 -point flexure
P/2 P,
(b) 4 -pointflxr
Fig. A 2 Loading configurations in flexure tests
-73-
of the equivalent cross section which becomes an "I" beam, is given by
bjh1 3 - (I-n) b2hz)12
Then, for a three point flexure (center-point loading) test (Figure A-2)
3P lh 1TRS - 2{blh1 - (1-n) b';,h23 }
It is assumed in the above equation that the failure occurs in the center
of the span. When the failure occurs elsewhere, the flexure formula is
modified to
TRS 3Ph (1- 2x)
2jblh, - (1-n) bh,
Where x is the distance between the central loading pin and the point of
fracture. The expression for the four-point flexure test is
TRS (4-point) = b h- (-n) b 2 h 23
where a can be /3 (1/3 - four point) or Z/4 (1/4 - four point), as shown
in Figure A-2,
-
i~i'.-74-
DISTRIBUTION LIST
No. of Copies
Naval Air Systems Command 39
Washington, D.C. 20361Attention: AIR OOD4 12
310C 1320A I
5163D4 25
Office of Naval Research 1
Washington, D.C. 20360
Attention: Code 471
White Oak Laboratory 1
Naval Surface Weapons CenterWhite Oak, Maryland 20910
Attention: Code 2301
Naval Research Laboratory 1Washington, D.C. 20390
Attention: Code 6360
David W. Taylor Naval Ship Research & Development Center 1Annapolis, Maryland 21402
Attention: W. Smith, Code 2832
Naval Air Propulsion Center 1
Trenton, New Jersey 08638
Attention: R. Valori, Code PE 72
Naval Undersea Center 1
San Diego, California 92132
Attention: Dr. J. Stachiw
Naval Air Development Center 1
Materials Application Branch, Code 6061
Warminster, Pennsylvania 18974
Air Force Wright Aeronautical Laboratories 4
Wright-Patterson, Air Force Base
Dayton, OH 45433Attention: Dr. J. Dill POSL 1
Dr. H. Graham MLLM 1Mr. B.D. McConnell MLBT 1Ms. K. Lark MLTM 1
Brookhaven National Laboratory
Upton, N.Y. 11973
Attention: Dr. D. Van Rooyen
1I
I ,, , , ,1
No. of Copies
DirectorApplied Technology LaboratoryU.S. Army Research & Technology LaboratoriesFort Eustis, Virginia 23604Attntion: DAVDL-ATL-ATP (Mr. Pauze)
U.S. Army Research OfficeBox CM, Duke StationDurham, North Carolina 27706Attention: CRDARD
U.S. Army MERDCFort Belvoir, Virginia 22060Attention: W. McGovern (SMEFB-EP
Army Materials and Mechanics Research CenterWatertown, Massachusetts 02172Attention: Dr. R.N. Katz
NASA HeadquartersWashington, D.C. 20546Attention: J.J. Gangler, RRM
NASA Lewis Research Center 221000 Brookpark RoadCleveland, Ohio 44135Attention: Dr. E. Zaretsky
W. A. Sanders (49-1)and Dr. T. Hergell
Defense Advanced Research Project Office 2
1400 Weilson BoulevardArlington, Virginia 22209Attention: Dr. Van Reuth 1
Dr. Buckley 1
Inorganic Materials DivisionInstitute for Materials ResearchNational Bureau of StandardsWashington, D.C. 20234
University of CaliforniaLawrence Berkeley LaboratoryHearsy Mining BuildingBerkeley, California 94720Attention: Dr. L. Froschauer
i
No. of Copies
Department of EngineeringUniversity of CaliforniaLos Angeles, California 90024Attention: Profs. J.W. Knapp and G. Sines
Department of MetallurgyCase-Western Reserve UniversityCleveland, Ohio 44106Attention: Dr. A. Heuer
Engineering Experiment StationGeorgia Institute of TechnologyAtlanta, Georgia 30332Attention: J.D. Walton
Department of Engineering Research 1North Carolina State University
Raleigh, North Carolina 27606Attention: Dr. H. Palmour
Materials Research Laboratory 1Pennsylvania State UniversityUniversity Park, Pennsylvania 16802Attention: Prof. Rustum Roy
Rensselaer Polytechnic Institute 1110 Eighth StreetTroy, New York 12181Attention: R.J. Diefendorf
School of Ceramics 1Rutgers, The State UniversityNew Brunswick, New Jersey 08903
Virginia Polytechnic Institute 1Minerals EngineeringBlacksburg, Virginia 24060Attention: Dr. D.P. H. Hasselman
Aerospace Corporation 1Materials Laboratory
P.O. Box 95085Los Angeles, California 90045
Supervisor, Materials Engineering 1Department 93-39MAiResearch Manufacturing Company of Arizona
402 South 36th StreetPhoenix, Arizona 85034
*1 3
No. of Copies
Materials Development CenterAVCO System DivisionWilmington, Massachusetts 01887Attention: Tom Vasilos
Battelle Memorial InstituteCeramics Department505 King AvenueColumbus, Ohio 43201
Metals and Ceramics Information CenterBattelle Memorial Institute505 King AvenueColumbus, Ohio 43201
Research and Development DivisionCarborundum CompanyNiagara Falls, New York 14302Attention: Mr. C. McMurty
Ceramic Finishing CompanyBox 498State College, Pennsylvania 16801
Ceradyne, Inc. 1Box 1103Santa Ana, California 92705
Coors Porcelain Company600 Ninth StreetGolden, Colorado 80401Attention: Research Department
* Federal-Mogul CorporationAnti-Friction Bearing R&D Center3980 Research Park DriveAnn Arbor, Michigan 48104Attention: D. Glover
Metallurgy and Ceramics Research DepartmentGeneral Electric R&D LaboratoriesP.O. Box 8Schenectady, New York 12301
Space Sciences Laboratory
General Electric CompanyP.O. Box 8555Philadelphia, Pennsylvania 19101
4
No. of Copies
Hughes Aircraft CompanyCulver City, California 90230Attention: Mr. M.N. Gardos
IIT Research Institute10 West 35th Street
Chicago, Illinois 60616Attention: Ceramics Division
Industrial Tectonics, Inc.18301 Santa Fe AvenueCompton, California 90224Attention: Hans R. Signer
Kaweki-Bervlco IndustryBox 1462
Reading, Pennsylvania 19603Attention: Mr. R.J. Longnecker
Research and Development DivisionArthur D. Little CompanyAcorn ParkCambridge, Massachusetts 02140
Mechanical Technology, Inc.968 Albany-Shaker RoadLatham, New York 12110Attention: Dr. E.F. Finkin
North American Rockwell Sicence CenterP.O. Box 1085Thousand Oaks, California 91360
Norton CompanyIndustrial Ceramics DivisionOne New Bond Street
Worcester, Massachusetts 01606Attention: Dr. M. Torti
Ceramic Division:4 Sandia Corporation
Albuquerque, New Mexico 87101
Engineering and Research Center
SKF Industries, Inc.1100 First AvenueKing of Prussia, Pennsylvania 19400Attention: L. Sibley
Solar Turbines InternationalP.O. Box 80966San Diego, California 92138Attention: G.W. Hosang
.7
2
No. of Copies
Southwest Research Institute 1P.O. Drawer 28510
San Antonio, Texas 78228
Materials Sciences & Engineering Laboratory 1Stanford Research InstituteMenlo Park, California 84025Attention: Dr. Cubiciotti
Teledyne CAE1330 Laskey RoadToledo, Ohio 43601Attention: Hugh Gaylord
Union Carbide CorporationParma Technical CenterP.O. Box 6116Cleveland, Ohio 44101
Materials Sciences LaboratoryUnited Aircraft CorporationEast Hartford, Connecticut 06101
Attention: Dr. J.J. Brennan
Astronuclear Laboratory 1Westinghouse Electric CorporationBox 10864Pittsburgh, Pennsylvania 15236
Westinghouse Research LaboratoriesBeulah RoadChurchill BoroughPittsburgh, Pennsylvania 15235Attention: Dr. R. Bratton
6
r mC